archive-license mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Brian Behlendorf <>
Subject Re: New license and copyright dates
Date Wed, 04 Feb 2004 05:04:38 GMT
On Wed, 4 Feb 2004, Mahesh T. Pai wrote:
> Brian Behlendorf said on Mon, Feb 02, 2004 at 10:05:44PM -0800,:
>  > I personally  don't see  why a file  with "Copyright  1999-2004" is
>  > going to be materially more or less effective
> You are  disabled from saying `I  published that before  you did' when
> faced  wiah a  suit by  somebody saying  that your  code  infringes on
> theirs.

Uh, but the artifact being published was published in 2004, not 1998.
Simply putting "1998" into the license does nothing to prove I actually
wrote any of that code in 1998, and it certainly doesn't mean all of it
was available in 1998.  If faced with such a suit, we'd reach back into
the CVS history and/or and/or our own release packages in our
archive to show what was published when.  Older releases don't suddenly
vanish when a newer release comes out.

> > than having a CVS tree with time stamps going that far back.
> The most important record of publication of a document is the document
> itself,  here  the source  code.   Remember  the  copyright /  version
> histories somewhere in the first pages of a book?

This is somewhat academic until we've got cases decided in court to set a
precedent for what's an authoritative claim of date of authorship.  I know
that I'd rather go in to court with a URL to with a date
showing the release of Apache 0.9.1 in 1995 rather than the current code
showing "copyright 1995-2004".

Books aren't different in publishing their version history.  Most
software, commercial and open source, have changelogs or release notes
that detail what changes were made when.  But knowing when a specific
algorithm or snippet of code is best demonstrated by showing the CVS tree.

In fact, it might behoove us to make regular tape backups of the CVS tree
and submit them to a records firm like Iron Mountain, which can timestamp
them and store them in perpetuity in a way that is irrefutable.  I
understand some companies have done this with the copy of Linux they
downloaded from SCO under the GPL, just in case SCO's lawyers send them a
pay-up letter.

IANAL, TINLA, if it's convenient to keep the year ranges when updating the
license there's probably no harm in doing so.  But I don't see a need to
mandate it.  If others in the ASF disagree, please speak up, I don't want
to be the last word on this.

>  > For  the purposes of anyone trying to follow
>  > the terms of the license, the most recent date is all that matters.
> Terms of license, yes. But when it comes to copyright ownerhship and
> infringement claims, it is a different story.

And nothing you put in the license can change that story.  :)


View raw message